and Heavy-Duty Diesel - ACS Publications - American Chemical Society

Nov 11, 2013 - Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720-1710, United States. §. Enviro...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/est

Quantifying On-Road Emissions from Gasoline-Powered Motor Vehicles: Accounting for the Presence of Medium- and Heavy-Duty Diesel Trucks Timothy R. Dallmann,†,‡ Thomas W. Kirchstetter,†,§ Steven J. DeMartini,† and Robert A. Harley*,†,§ †

Department of Civil and Environmental Engineering, University of California, Berkeley, California 94720-1710, United States Environmental Energy Technologies Division Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States

§

S Supporting Information *

ABSTRACT: Vehicle emissions of nitrogen oxides (NOx), carbon monoxide (CO), fine particulate matter (PM2.5), organic aerosol (OA), and black carbon (BC) were measured at the Caldecott tunnel in the San Francisco Bay Area. Measurements were made in bore 2 of the tunnel, where light-duty (LD) vehicles accounted for >99% of total traffic and heavy-duty trucks were not allowed. Prior emission studies conducted in North America have often assumed that route- or weekend-specific prohibitions on heavy-duty truck traffic imply that diesel contributions to pollutant concentrations measured in on-road settings can be neglected. However, as light-duty vehicle emissions have declined, this assumption can lead to biased results, especially for pollutants such as NOx, OA, and BC, for which diesel-engine emission rates are high compared to corresponding values for gasoline engines. In this study, diesel vehicles (mostly medium-duty delivery trucks with two axles and six tires) accounted for 99% of total vehicle counts during the 4−6 p.m. commute period), so previous studies used eq 1 and pollutant measurements from bore 2 of the tunnel to determine emission factors for LD vehicles directly. An implicit assumption was that small numbers of MD and HD trucks driving through bore 2 did not contribute significantly to measured pollutant concentrations. If pollutant emission factors for LD vehicles 13874

dx.doi.org/10.1021/es402875u | Environ. Sci. Technol. 2013, 47, 13873−13881

Environmental Science & Technology

Article

and diesel trucks are comparable, then this assumption is valid. However, as LD vehicle emissions have decreased over time, emission factors for some species (e.g., PM2.5, BC, NOx) are now an order of magnitude higher for diesel trucks relative to LD gasoline vehicles.18 In such cases, the presence of even small numbers of MD and HD trucks in bore 2 of the tunnel might bias the determination of LD vehicle emission factors. Results from concurrent measurements of emission factors from individual diesel trucks at the Caldecott tunnel15 were used in this study for the apportionment of measured pollutant concentrations to determine contributions from LD gasoline vehicles and MD/HD diesel trucks. This approach explicitly accounts for the presence of diesel trucks in bore 2 of the tunnel and more accurately characterizes emission factors for LD vehicles. The data analysis methods applied here follow previous efforts to apportion emissions in mixed-use traffic lanes with much higher diesel truck fractions.5,18 In past studies at the Caldecott tunnel,18,25 apportionment efforts focused on deducing the diesel contribution to pollutant emissions in a mixed traffic bore of the tunnel, for midday sampling periods when diesel trucks were more abundant. Here, the goal was to determine contributions to pollution in bore 2 of the tunnel, for which MD/HD vehicle contributions were previously neglected. The relative contribution of gasoline vehicles (Δ[CO2]g) to measured tunnel CO2 concentrations was estimated using traffic count data and fuel and vehicle properties Δ[CO2 ]g Δ[CO2 ]

Δ[CO2 ]d = Δ[CO2 ] − Δ[CO2 ]g

The LD gasoline contribution to background-subtracted concentrations of other species measured in the tunnel (Δ[P]g) was then calculated as ⎛ Δ[P]HD ⎞ Δ[P]g = Δ[P] − Δ[CO2 ]d ⎜ ⎟ ⎝ Δ[CO2 ]HD ⎠

ρg wg[fLD ULD + fMD (1 − F )UMD,g ] ρg wg[fLD ULD + fMD (1 − F )UMD,g ] + ρd wd(fHD UHD + fMD FUMD,d)

(2)

where ρ and w are the fuel density and carbon weight fraction, respectively, for gasoline and diesel fuel (subscripts g and d, respectively); U is the fuel consumption (L/100 km) for each vehicle type; and F is the fraction of MD trucks powered by diesel engines. Values for these parameters are summarized in Table 1. f represents the fractions of observed vehicles that fall into the LD, MD, and HD categories; these values varied somewhat from day to day throughout the study. The contribution of MD and HD diesel trucks (Δ[CO2]d) to the measured CO2 enhancement was subsequently calculated as

⎛ ⎞ Δ[P]g ⎟⎟wg EFP,LD = ⎜⎜ ⎝ Δ[CO2 ]g + Δ[CO]g ⎠

units

value

L (100 km)−1 L (100 km)−1 L (100 km)−1 L (100 km)−1 kg L−1 kg L−1 kg of C (kg of fuel)−1 kg of C (kg of fuel)−1 −

10.3 28.4 27.0 49.5 0.742 0.852 0.824 0.866 0.7

(5)

Emission factors calculated using eq 5 explicitly account for the presence of MD and HD trucks in bore 2 and better represent actual emission factors for LD vehicles. For pollutants that were measured at high time resolution (i.e., BC, NOx, and CO), 1-h average data were used in eq 5 to provide two discrete emission factor results per day, yielding a larger sample of 16 1-h average values over the eight-day sampling campaign. For filter-based measurements of PM2.5 and OA, only one LD emission factor could be calculated for each of the 2-h sampling periods. Other carbon-containing species accounted for less than 1% of the total vehicle-derived carbon emissions measured in bore 2 and were thus excluded from the denominator of eq 5. In addition to the vehicle count apportionment method described above, a second method utilizing high-timeresolution data from one day of sampling (July 12) was used to investigate further the contribution of MD and HD trucks to measured pollutant concentrations in bore 2 of the tunnel. Times at which individual MD and HD trucks passed by the sampling inlet were identified from video recordings of tunnel traffic. For each truck, a corresponding exhaust plume was identified, where possible, in the 1 Hz BC, NOx, and CO2 concentration data. Periods with a high influence of diesel exhaust were often readily identifiable using BC concentration

Table 1. Vehicle and Fuel Parameters Used in Eq 2 ULDa UMD,ga UMD,da UHDa ρgb ρdb wgb wdb Fc

(4)

Here, the ratio in parentheses represents the emission ratio of pollutant P to CO2 for HD diesel trucks. With the exception of PM2.5, values for this ratio were derived from measurements of emission factors for individual HD diesel trucks conducted as part of this field campaign and reported elsewhere.15,26 Note that OA emission factors for HD diesel trucks were derived from measurements of individual truck exhaust plumes, which are less dilute than vehicle exhaust measured in bore 2 of the tunnel. The reduced dilution could enhance partitioning of semivolatile organic compounds in diesel exhaust to condensed phases and result in an overestimate of the HD diesel OA emission factor relative to dilution levels prevailing for LD vehicle emissions in bore 2 of the tunnel.28−30 Fine particulatematter emission factors for HD diesel trucks were not measured directly but were calculated here as the sum of BC and OA emission factors. Supporting measurements at the Caldecott tunnel indicate that these carbonaceous species account for greater than 90% of PM2.5 mass emissions in motor vehicle exhuast.26 In eq 4, HD truck emission factors are assumed to apply to both MD and HD diesel trucks. Exhaust emission standards, expressed in mass of pollutant emitted per unit of useful work output by the engine, for both MD and HD diesel trucks have historically been set at similar levels for the pollutants considered here, so similar emission factors are expected.31 Results from eqs 2 and 4 were then used to calculate adjusted pollutant emission factors (EFP,LD) for LD gasoline vehicles

=

parameter

(3)

a

Fuel consumption rates from Ban-Weiss et al.18 bFuel properties from analysis of gasoline and diesel fuel samples collected in California in summer 2010.37 cDiesel fraction of MD vehicles estimated using California Air Resources Board EMFAC2011 model data45 and calculated as weighted average by fuel consumption of LHDT2 and MHD vehicle categories. 13875

dx.doi.org/10.1021/es402875u | Environ. Sci. Technol. 2013, 47, 13873−13881

Environmental Science & Technology

Article

to isolate the emissions signal from vehicles traveling through the tunnel. Contributions from individual MD and HD trucks to measured pollutant concentrations are shown in blue, whereas green shading denotes contributions from much larger numbers of LD vehicles. During this 2-h sampling period, 20 MD and 3 HD trucks were noted driving through bore 2 of the tunnel. Although MD and HD trucks accounted for less than 1% of total vehicle counts, their contributions to pollutant concentrations are clearly noticeable in Figure 1. The impact of diesel truck emissions is most clearly highlighted in the data recorded at 16:46, when an HD truck drove through the tunnel. Sharp increases in pollutant concentrations above typical tunnel levels are apparent in Figure 1 and correspond to the time period when the exhaust plume from this truck was being sampled. The peak BC concentration was nearly 100 times higher than typical tunnel concentrations, whereas smaller increases (factors of 6 and 2, respectively) were observed for NOx and CO2. This single HD diesel truck accounted for 19% of the total BC measured during the entire 2-h sampling period and 2% of the total NOx signal. When normalized to fuel consumption, BC and NOx emission factors for this truck were found to be similar to fleet-average values for HD trucks reported by Dallmann et al.15 Thus, this truck was not an especially high emitter relative to the HD truck population at large. Rather, the disproportionate contributions to measured BC and NOx are indicative of the large differences in emission factors for these pollutants from HD diesel trucks when compared to those for LD gasoline vehicles. Another contributing factor might be preferential sampling of emissions from trucks with elevated exhaust stacks, given the location of the air sampling inlets in the tunnel ventilation system above the traffic lanes. In general, the presence of diesel trucks had the most pronounced effect on BC concentrations, accounting for 40% of the total BC measured during the 2-h sampling period (Figure 1). A similar response to MD and HD trucks was observed for NOx, although concentration increases observed for truck exhaust plume events were not as large as those for BC. For this sampling period, diesel trucks were estimated to contribute 11% of total NOx emissions. In contrast to BC and NOx, CO2 concentrations in exhaust plumes for most trucks were not noticeably different from the baseline tunnel concentrations. Light-duty vehicles accounted for nearly all of the CO 2 measured in the tunnel, with minor (2%) contributions from MD and HD trucks. In this case, the attributed CO2 concentration was approximately proportional to the relative fraction of each vehicle type present in the tunnel. These results demonstrate the substantial impact that relatively small numbers of MD and HD trucks can have on time-averaged concentrations of air pollutants measured in a mixed-use roadway setting. That diesel engines have considerably higher NOx and BC emission factors than gasoline engines is not surprising and has been documented extensively elsewhere.17,18,22,25 However, the data reported here indicate that these emission differences are of such a magnitude that even small numbers of diesel trucks can contribute substantially to overall emissions of these pollutants, even in cases that appear to be heavily dominated by LD vehicle traffic. LD Vehicle Emission Factors. The relative contribution of diesel trucks to measured species concentrations in bore 2 was assessed over each of the eight sampling days using eqs 2−4. Here, the apportionment of measured species concentrations in

data, for which clear increases in BC concentration above tunnel baseline levels corresponded to the passage of individual trucks. These plume events were integrated and backgroundcorrected for each pollutant and were compared against the entire 2-h sampling period total to infer the relative contributions from diesel trucks and LD vehicles. Lowertime-resolution measurements of other species did not permit use of this alternative analytical approach for CO, PM2.5, or OA. Results from this additional apportionment method were compared to apportioned emissions evaluated based on observed vehicle counts for the July 12 sampling period.



RESULTS AND DISCUSSION Vehicle Activity. For the weekday 4−6 p.m. sampling periods considered here, LD vehicles accounted for greater than 99% of total vehicles observed, averaging 3625 veh hr−1. Of the trucks observed in bore 2, most were medium-duty (two-axle/ six-tire) vehicles, observed at an average rate of 23 veh hr−1. Heavy-duty trucks were observed in bore 2 on only four of the eight days of sampling, with a maximum influence on July 14 when eight HD trucks were identified during the 2-h sampling period. Light-duty vehicle activity in the tunnel was stable from day to day, with little variation in total vehicle counts observed across sampling days. Traffic was relatively free-flowing once vehicles entered the tunnel, at typical speeds of 60 km hr−1 on a 4% uphill grade. Measurements reported here reflect emissions from vehicles operating in warmed up, stabilized modes. Excess emissions associated with cold engine starting, which can be an important contributor to total emissions from LD gasoline vehicles,32,33 were not measured in this study. Influence of Individual Truck Exhaust Plumes. Timeseries plots of tunnel BC, NOx, and CO2 concentrations measured on July 12 are shown in Figure 1. For each species, corresponding background air concentrations were subtracted

Figure 1. Ambient-background-subtracted tunnel concentrations of CO2 (bottom), NOx (middle), and BC (top) during the 7/12/2010 sampling period apportioned between LD vehicles (green) and MD and HD trucks (blue). Apportionment is based on an analysis of video recording of tunnel traffic and identification of passing times for individual MD and HD trucks. Inset pie charts show relative contributions of each vehicle type to measured pollutant concentrations for the 2-h sampling period. Average ambient concentrations of CO2, NOx, and BC were 394 ppm, 0.024 ppm, and 0.7 μg m−3, respectively. 13876

dx.doi.org/10.1021/es402875u | Environ. Sci. Technol. 2013, 47, 13873−13881

Environmental Science & Technology

Article

the tunnel was based on measured vehicle counts and fleetaverage pollutant emission factors for HD trucks. Backgroundsubtracted tunnel concentrations for each sampling period are shown in Figure 2, with calculated contributions from MD and

Table 2. Unadjusted and Adjusted Emission Factors for LD Gasoline Vehicles pollutant

N

unadjusted emission factor (EFP)a (g kg−1)

NOxb CO PM2.5 BCc OA

16 16 8 16 8

2.29 ± 0.12 14.2 ± 0.7 0.048 ± 0.012 0.018 ± 0.002 0.021 ± 0.006

adjusted emission factor (EFP,LD)a (g kg−1)

relative changea (%)

1.90 ± 0.08 14.3 ± 0.7 0.038 ± 0.010 0.010 ± 0.002 0.017 ± 0.005

−17 ± 5 +1 ± 7 −22 ± 28 −43 ± 14 −20 ± 34

a

Unadjusted values calculated using eq 1; adjusted values calculated using eq 5 to exclude pollutant contributions from MD and HD diesel trucks present in bore 2 of the tunnel. Uncertainty estimates provide 95% confidence intervals. bNOx emission factors reported as NO2 equivalents. cBC emission factors calculated using Aethalometer BC concentration data for tunnel measurements and filter-derived BC concentration data for ambient-background subtractions.

calculated without accounting for diesel truck contributions using eq 1. With the exception of CO, LD vehicle emission factors were all found to be lower than the unadjusted values calculated using eq 1. The largest relative effect was seen for the BC emission factor, which was reduced by (43 ± 13)% after accounting for diesel truck contributions. Likewise, the adjusted PM2.5 and OA emission factors decreased relative to unadjusted values, although these changes have larger associated uncertainties. The reduction in the NOx emission factor for LD vehicles was smaller, although still significant. Diesel trucks were an insignificant source of CO when compared to emissions from the large numbers of gasoline vehicles in the tunnel, and the adjusted CO emission factor was found to be similar to the unadjusted value. For all species considered here, LD vehicle emission factors were lower than those measured at the same location in 2006.14,18 These reductions resulted from both continued longterm downward trends in emissions from the LD vehicle fleet as new, lower-emitting vehicles replace older vehicles and the new methods for calculating emission factors used here that address a positive bias in previous estimates of LD NOx, PM2.5, BC, and OA emission factors. The light-duty NOx and CO emission factors reported here are, respectively, 54% and 21% lower than the values measured in a 2010 remote-sensing study of LD vehicles in Los Angeles, CA.10 For NOx, this relation is consistent with previous comparisons of LD emission factors measured at the Caldecott tunnel and remote-sensing measurements, with differences in emission factors likely influenced by vehicle fleet age and operating mode differences.10,34,35 The mean LD vehicle fleet age at the Caldecott tunnel has historically been approximately 3 years lower than the mean fleet age for vehicles operating at the Van Nuys tunnel in Los Angeles.10,35 The mean vehicle fleet age at the Caldecott tunnel was 6.3 years in 2006,18 although the fleet in 2010 might have been slightly older because of the effects of the economic recession. The relatively new vehicle fleet likely contributes to the lower NOx and CO emission factors observed at the Caldecott tunnel. After adjustments to account for diesel-engine contributions to tunnel concentrations measured in this study, there are large differences in resulting BC emission factors when compared to emission factors from other North American cities. The BC emission factor for LD vehicles estimated here is 7 times lower than the mean BC emission factor derived from on-road measurements in Los Angeles20 and 11 times lower than the

Figure 2. Measured increase in species concentration above ambient levels apportioned to LD gasoline vehicles (green) and HD and MD trucks (blue). Hourly time-resolved data were not available for OA and PM2.5; therefore, 2-h average values for these species are presented here.

HD trucks shown in blue and LD vehicle contributions shown in green. Particle-phase emissions were the most sensitive to the presence of diesel trucks in the vehicle mix. The mean (±95% confidence interval) diesel contributions to measured PM2.5, OA, and BC concentrations were (24 ± 4)%, (22 ± 6)%, and (45 ± 8)%, respectively. The two days with the highest levels of diesel truck activity, July 8 and July 14, also had the highest measured tunnel concentrations of OA and PM2.5. Similarly, relatively high levels of NOx were also measured on these days. The average diesel contribution to NO x concentrations was (18 ± 3)%. Emissions of CO and CO2 were dominated by LD vehicles, with diesel contributions of less than 2%. A comparison of the two apportionment methods for July 12 shows generally good agreement in the apportionment of emissions between vehicle types. Both methods attributed 37−40% of BC and 1−2% of CO2 emissions to diesel trucks. The exhaust plume analysis estimated a lower NOx contribution from trucks (11%) relative to the vehicle count apportionment method (17%), possibly because the HD-truckderived emission factor might overestimate actual NO x emission factors for MD trucks. Once vehicle contributions to tunnel pollutant concentrations had been quantified, LD vehicle emission factors were calculated using eq 5. The results are summarized in Table 2 and compared with unadjusted emission factors that were 13877

dx.doi.org/10.1021/es402875u | Environ. Sci. Technol. 2013, 47, 13873−13881

Environmental Science & Technology

Article

LD vehicle emission factor inferred from near-road measurements along a highway running north of Toronto.5 In both of these studies, emissions measurements were made for vehicles traveling on highways, and operating modes for the sampled vehicle population did not differ substantially from those observed at the Caldecott tunnel. The differences in BC likely arise from the inherent difficulties in characterizing emission factors based on measurements on or near mixed-use roadways. In the Los Angeles study, LD emission factors were calculated from measurements made from a highway with